1. Overview
The main purpose of this assignment is to evaluate the take-home exercise 1 from the perspective of clarity and aesthetics, and to refine the diagram based on the principles we learned in the previous two lessons. The dataset of this project is VAST Challenge 2022. The operation was carried out on Rstudio and main packages used are tidyverse and ggplot2 extensions.
2. Data Preparation
We will do the data preparation in a similar procedure of the original creation:
Installing and loading the required libraries
Importing the dataset
hide
# read csv file
participants <- read_csv('data/Participants.csv')
Data Wrangling
hide
# rename 'HighSchoolOrCollege'
participants$educationLevel <- sub('HighSchoolOrCollege',
'High School or College',
participants$educationLevel)
# rename columns
participants <- participants %>%
rename('ParticipantID' = 'participantId',
'HouseholdSize' = 'householdSize',
'HaveKids' = 'haveKids',
'Age' = 'age',
'EducationLevel' = 'educationLevel',
'InterestGroup' = 'interestGroup',
'Joviality' = 'joviality')
# Age variable is binned with the following code chunk:
brks <- c(17, 20, 25, 30, 35, 40, 45, 50, 55, 60)
grps <- c('20 & Below', '21-25', '26-30', '31-35', '36-40', '41-45','46-50', '51-55', '56-60')
participants$AgeGroup <- cut(participants$Age, breaks=brks, labels = grps)
3. Critique and New Design
3.1 Distribution of Age
A histogram is used to reveal the distribution of residents’ age.
hide
knitr::include_graphics("C:/KatherineHuan/ISSS608/th_ex/data/pic/th2_or1.PNG")
3.1.1 Critique
Clarity
- In general,this chart gives a rough idea of the age distribution of the participants, but it more information can be extracted, such as the percentage of each age group
- Annotation of counts of each age group conveys intuitive information for the audience .
- The superscript of each bar chart is not intuitive and it is better to use percentages.
- The x-axis title format looks not formal.
- The meaning of y-axis title is vague, can be improved to be more specific.
- The position of x-axis tick mark is not easy for reading.
- Adding more data would be better,such as data sources.
Aesthetic
- The color contrast is not beautiful, a lighter color for the background and a brighter color for the bar will be better.
- The color of the annotation is not harmonious.
- The title is not in the middle of the figure, affecting the appearance.
- Intuitive bar chart will be more attractive and informative.
3.1.2 New design
hide
library(plotly)
# Bar chart plot
Age <- ggplot(data= participants,
aes(x= AgeGroup)) +
geom_bar(fill= 'light blue') +
ylim(0, 150) +
# annotation
geom_text(stat = 'count',
aes(label= paste0(round(stat(count)/sum(stat(count))*100,
1), '%')), vjust= -0.5, size= 2.5) +
labs(y= 'No. of\nResidents', x= 'AgeGroup',
title = "Distribution of Residents' Age",
subtitle= 'Demographics in Engagement, Ohio',
caption = "Source: VAST Challenge 2022") +
theme(plot.title = element_text(hjust = 0.5),
plot.subtitle = element_text(hjust = 0.5),
axis.title.y= element_text(hjust = 0.5),
axis.ticks.x= element_blank(),
panel.background= element_blank(),
panel.grid.major = element_line(size= 0.2, color = "grey"),
axis.line= element_line(color= 'grey'),
plot.caption = element_text(hjust = 0))
ggplotly(Age)
Clarity
- The residents’ proportion in each age group was shown, which is more intuitive for audience to understand.
- Data source and explanation of the data are added to provide more information.
- The Intuitive effect helps audience to explore more detailed information when they are interested.
- The mark of x-axis and y-axis is easy to understand.
Aesthetic
- The color match is clean and easily readable to the audience.
- Positions of title, y-axis are improved
- The plot backgroud is clearer
3.2 Joviality Distribution in Relation to AgeGroup and EducationLevel
Line chart is used to reveal the distribution of residents’ joviality according to their age and education level.
hide
knitr::include_graphics("C:/KatherineHuan/ISSS608/th_ex/data/pic/th2_or2.PNG")
3.2.1 Critique
Clarity
- There are three variables in this picture, age, education level and Joviality, which is intended to show the joviality of different ages and education levels.With too many factors and no final focus, it is difficult to express the author’s meaning clearly.
- There are too many lines and multiple colors, which make it looks complicated.It is difficult to see the distribution trend of Joviality of different ages on each line.
- The lines are too thick, and the distribution curves of different education levels cover each other, so it is difficult to see the average level of joviality of different education levels in each age group.
- The title on the X-axis seems ambiguous and informal.
- The gray background distracts the focus and makes it look indistinct.
- Additional information, such as data sources, can be added to make the icon more complete.
- The font design makes the theme unclear and unfocused.
Aesthetic
- The Y-axis title looks informal; it could be more formal.
- The title is not centered and can be adjusted to look better.
- The size of the annotation doesn’t look pretty.
- The overall color scheme lacks design and is not attractive.
3.2.2 New design
a. Joviality in relation to Age and Education Level
hide
# Plot Joviality in relation to Age and Education Level
p1 <- ggplot(participants,
aes(x= AgeGroup, y= Joviality)) +
geom_violin(fill= '#66cdaa',
scale = 'count',alpha = .9, trim = FALSE,
color= NA) +
geom_boxplot(width= 0.2,
color = '#065535',
alpha= 0.8) +
stat_summary(aes(color= 'Mean'),
fun= 'mean',
size= 0.05) +
geom_hline(aes(yintercept = 0.5),
color= 'black',
linetype= 'dashed',
size= .6) +
ylim(-0.3,1.5) +
# Annotation
scale_color_manual(name= 'Statistics',
values = (Mean= '#f6546a')) +
# Add on Education Level Factor
facet_grid(~factor(EducationLevel)) +
labs(title= 'Joviality Distribution in Relation to Age',
x= 'Age Group', y= 'Joviality',
subtitle= 'Demographics in Engagement, Ohio',
caption = "Source: VAST Challenge 2022") +
theme(panel.background = element_blank(),
plot.title = element_text(hjust = 0.5),
plot.subtitle = element_text(hjust = 0.5),
axis.title.y = element_text(angle=0,vjust = 0.5),
axis.ticks.x = element_blank(),
axis.line= element_line(color= 'grey'),
axis.text.x = element_text(size=8,angle=90),
panel.grid.major.y = element_line(color= 'grey', size = 0.1),
plot.caption = element_text(hjust=0),
legend.key = element_rect(fill= NA))
p1
To detect the relationship between every two variables, we will plot another 3 charts for better understand how is the incluence in between.
b. Joviality in relation to Age
hide
# Plot Joviality in relation to Age
p2 <- ggplot(participants,
aes(x= AgeGroup, y= Joviality)) +
geom_violin(fill= '#66cdaa',
scale = 'count',alpha = .9, trim = FALSE,
color= NA) +
geom_boxplot(width= 0.2,
color = '#065535',
alpha= 0.8) +
stat_summary(aes(color= 'Mean'),
fun= 'mean',
size= 0.2) +
geom_hline(aes(yintercept = 0.5),
color= 'black',
linetype= 'dashed',
size= .6) +
# Annotation
scale_color_manual(name= 'Statistics',
values = (Mean= '#f6546a')) +
labs(title= 'Joviality Distribution in Relation to Age',
x= 'Age Group', y= 'Joviality',
subtitle= 'Demographics in Engagement, Ohio',
caption = "Source: VAST Challenge 2022") +
theme(panel.background = element_blank(),
plot.title = element_text(hjust = 0.5),
plot.subtitle = element_text(hjust = 0.5),
axis.title.y = element_text(angle=0,vjust = 0.5),
axis.ticks.x = element_blank(),
axis.line= element_line(color= 'grey'),
axis.text.x = element_text(size=8,angle=0),
panel.grid.major.y = element_line(color= 'grey', size = 0.1),
plot.caption = element_text(hjust=0),
legend.key = element_rect(fill= NA))
p2
c. Joviality in relation to Education Level
hide
# Plot Joviality in relation to Education Level
p3 <- ggplot(participants,
aes(x= EducationLevel, y= Joviality)) +
geom_violin(fill= '#66cdaa',
scale = 'count',alpha = .9, trim = FALSE,
color= NA) +
geom_boxplot(width= 0.08,
color = '#065535',
alpha= 0.8) +
stat_summary(aes(color= 'Mean'),
fun= 'mean',
size= 0.2) +
geom_hline(aes(yintercept = 0.5),
color= 'black',
linetype= 'dashed',
size= .6) +
# Annotation
scale_color_manual(name= 'Statistics',
values = (Mean= '#f6546a')) +
# Add on Education Level Factor
#facet_grid(~factor(AgeGroup)) +
labs(title= 'Joviality Distribution in Relation to Education Level',
x= 'Education Level', y= 'Joviality',
subtitle= 'Demographics in Engagement, Ohio',
caption = "Source: VAST Challenge 2022") +
theme(panel.background = element_blank(),
plot.title = element_text(hjust = 0.5),
plot.subtitle = element_text(hjust = 0.5),
axis.title.y = element_text(angle=0,vjust = 0.5),
axis.ticks.x = element_blank(),
axis.line= element_line(color= 'grey'),
axis.text.x = element_text(size=8,angle=0),
panel.grid.major.y = element_line(color= 'grey', size = 0.1),
plot.caption = element_text(hjust=0),
legend.key = element_rect(fill= NA))
p3
d. Education Level in relation to Age
hide
# Data manipulation
percent <-
participants %>%
group_by(AgeGroup, EducationLevel) %>%
summarise(edu_size= n()) %>%
mutate(edu_pct= percent(edu_size/sum(edu_size)))
#Plot Education Level in relation to Age
p4 <- ggplot(data= percent,
aes(x= factor(AgeGroup),
y= edu_pct,
group= EducationLevel,
color= factor(EducationLevel))) +
geom_line() +
scale_color_discrete(name= 'Education Level') +
scale_y_continuous(labels = percent_format(),
expand = c(0.2, 0.2)) +
labs(x= 'Age Group',y= 'Propotions',
title = "Composition of Residents' Education Level in Age Groups")+
theme(legend.position = 'top', legend.direction = 'horizontal',
axis.title.y= element_text(angle=90), axis.ticks.x= element_blank(),
panel.background= element_blank(), axis.line= element_line(color= 'grey'),
panel.grid.major = element_line(size= 0.2, color = "grey"),
legend.key = element_rect(fill= NA), legend.title = element_text(size = 8.5),
plot.title = element_text(hjust=0.5))
p4
Clarity
- The new graph adopted a combination of violin, box plot and line chart, instead of single line chart to reveal the distribution of joviality for residents.
- With ordinary density curves, it is difficult to compare more than just a few distributions because the lines visually interfere with each other. With a violin plot, it’s easier to compare several distributions since they’re placed side by side. So here we used a combined violin plot to have a overall visualization of the relationship of variables.
- Three individual charts are adopted to show the influence of age and education level on Joviality and influence of age on education level respectively, which is easier to detect the trend when considering each aspect.
- In the combined chart, the green area of each violin shows the density of joviality based on age group and education level, which intuitively reflect the joviality level. And the facet of education level makes it possible to take the third factor into consideration when explain the relationships to the audience.
- Data source and explanation of the data are added to provide more information.
- The ticks mark of x-axis and y-axis are easy to understand.
- The color match is clean and easily interpretable to the audience.
- The legend make the chart more readable and easier to understand.
- The horizontal line showing the mean of joviality make it easier to compare and interprete the data.
- The size of the red point indicating the mean joviality in the box plot make the data is easier to read and interpret.
Aesthetic
- Positions of title, y-axis are improved
- The plot background is clearer and easier to detect changes.
- Overall format and design are improved to provide a clean and professional look.
4. Conclusions
There are more than one way to present the data.It’s not that the more complex the graph, the better to show the ability to make the graph, we should avoid pack too much information into one graph, but use simple and clear graph to make the audience understand our point of view more easily.
Derived values reveal more interesting patterns than absolute values.For example, the proportion of each age group. So, pre-attentive processing is important for design of visualizations such as the calculation of proportion of education level.